Skip to content

Automations

An automation is a small flow you build on a canvas : a trigger ("when this happens") wired to one or more actions ("do these things"). When the trigger fires, the flow runs on the server, on its own ; you don't have to be watching.

Reach it from the Automation entry in the navigation rail (visible where the feature is enabled). This page covers building and running a flow ; for the details of passing values between steps, see Using values in an automation.

Your automations

/automation lists the flows in your organization. Each has an Enabled switch ; a flow only fires automatically when it's on, so you can build and test one safely while it's off. New opens the editor on a blank canvas. You can search, open, duplicate-by-editing, archive, and restore flows.

Building a flow

The editor is a canvas. You drag nodes from the palette onto it and wire them together ; clicking a node opens its config panel on the side.

Every flow has exactly one trigger node at the start, which decides how it fires :

  • Event : fires when a chosen platform event happens (a user signs in, a record is created, a role is assigned, …). Pick the event from a grouped list ; the editor shows what data that event carries so you can use it downstream.
  • Manual : never fires on its own ; you run it yourself with Run now. Good for testing or on-demand jobs.
  • HTTP : fires when an external system sends a web request to the flow's URL, with the flow's secret. Use it to let another tool kick off a flow.
  • Schedule : fires on a recurring schedule you set in the schedule builder (every N minutes/hours, daily, weekly, or monthly), in your own timezone.

Wired after the trigger, action nodes do the work. What's available includes :

  • Communication : send an in-app / email notification, send an email, or call an outbound webhook.
  • Data : create / update / delete / find records in a Data Model.
  • People : assign or remove a role, get a user, update a user's profile or metadata, activate / deactivate a user.
  • Control flow : a condition that branches (true / false), a delay that pauses the flow (even for hours or days), set variable / constant / transform for shaping values.

Nodes pass values down the flow : a later node can reference the trigger's data or an earlier step's output with {{ … }} references, inserted by the variable picker. That's the subject of Using values in an automation.

Running a flow

  • Save to persist your edits. The editor checks the graph first and tells you if something's incomplete (no trigger, a required field left empty, a loop, …) before it will run.
  • Run now runs the flow once with a sample trigger payload, so you can watch it work without waiting for a real event.
  • Turn on Enabled for the flow to start firing from its trigger for real.

Run history & logs

Each flow keeps a run history : every run, when it happened, and whether it succeeded, is still running, or failed. Open a run to see it step by step ; each node's inputs, its output, and the log lines the node recorded. When a run fails, the history is where you see which node stopped it and why. Failed runs retry a few times automatically before they're marked failed.

What a flow is allowed to do

A flow runs as the person who created it, and it can never do more than that person could do by hand : the privileged actions (writing records, changing roles, managing users) re-check the owner's permissions every time they run, and shrink automatically if the owner loses a permission. So an automation is a convenience, not a way around your organization's access rules.

Reaching external systems safely

Action nodes that call out (webhook, email, or anything needing a token) read credentials from your organization's Secrets by name ; you reference a secret in the node's config, and the value is fetched at run time without ever being shown in the editor or the logs.